Agents-playground not displaying text in Speech to text agent

Agents
,
1

I want to create a simple Speech-to-Text agent that I can interact with on agents-playground, but I cannot get it to display the text in the “Chat” window, despite many hours debugging with Claude. The funny thing is, I did get the following code to work once, so seems the code doesn’t work consistently. Below is my code. Are there any way I can tweak it to make it work?

Version info:

livekit 1.1.2
livekit-agents 1.4.3
livekit-api 1.1.0
livekit-protocol 1.1.2

agents-playground: 59c5300

#!/usr/bin/env python3
import logging
import os

from livekit import agents
from livekit.agents import AgentServer, AgentSession, Agent, room_io

STT_GREETING = os.getenv("STT_GREETING", "STT agent is ready. Speak to transcribe.")

class Assistant(Agent):
    def __init__(self) -> None:
        super().__init__(
            instructions="You transcribe incoming audio using Azure Speech STT."
        )


server = AgentServer()


@server.rtc_session()
async def my_agent(ctx: agents.JobContext):
    # Configure STT with either endpoint or region

    session = AgentSession(
        stt=azure.STT(...),
    )

    await session.start(
        room=ctx.room,
        agent=Assistant(),
        room_options=room_io.RoomOptions(
            audio_input=True,
            audio_output=False,  # No TTS/audio output
            text_input=False,
            # Use TextOutputOptions instead of AudioOutputOptions
            text_output=room_io.TextOutputOptions(),
        ),
    )

    if STT_GREETING:
        print(STT_GREETING)


if __name__ == "__main__":
    agents.cli.run_app(server)

I ran python script.py dev and the transcription did appear in the console, but just not in Playground. Any advice?

Many thanks for your help!

2

TTT transcriptions are automatically published to the frontend/playground in realtime when an STT component is part of the session pipeline. No special output config is needed. I would suggest using this code as a starting point: agents/examples/other/transcription/multi-user-transcriber.py at main · livekit/agents · GitHub

3

I agree with Raghu, but I also wondered if there was some issue with your code, so I gave it a try. It worked for me without modification (obviously after populating the Azure STT params)

- uv run src/agent.py dev
- Navigate to agents-playground.livekit.io & select project associated with the agent's LiveKit URL/keys/secret
- 'Connect' & speak
- Transcribed text appears in chat window